Complex FFT for real data
I use a code based on Numerical recipes implementation of complex FFT. I work with real data. In my work, I do not need to use the FFT heavily, and therefore, I do not wish to invest a lot of time for increasing efficiency of solution or implementing other code.
To do 1-D forward FT of real data, I add zeros to the imaginary components and add zeros to both real and imaginary components until the vector length is N=2^n. After the transform, I consider data from 0 to N/2.
My problem is the IFFT. I pad zeros until N = 2^n and make sure that values are conjugate mirrored about the vector center. That is: Value[N-n-1] = conj Value[n].
After I do the IFFT, I expect the result to be only real values (img components should be zeros more or less). Nevertheless, this is not what happens. The imaginary components have values as if it were a normal complex IFFT.
Any suggestions are appreciated.
|All times are GMT -4. The time now is 18:22.|